Although neovascularization is regarded as an essential factor for tumor growth, it is unclear whether pancreatic adenocarcinoma is also influenced by this process. Furthermore, the reported microvessel count (MVC) data can not be compared due to the diversity of evaluating methods, and the relation between MVC data and metastatic potentials remains controversial. A total of 24 pancreatic adenocarcinomas and 24 adjacent non-cancerous pancreatic parenchyma were analyzed for MVC using anti-CD31 antibody. In addition, the MVC of 15 hypervascular tumors (10 hepatocellular carcinomas: HCC and 5 islet cell pancreatic tumors: ICT), 30 other types of adenocarcinomas (10 gastric, 10 colon and 10 intraductal papillary mucinous tumors of the pancreas: IPMT), as well as that of non-cancerous areas, were also analyzed. The extent of hepatic and peritoneal spread in 24 pancreatic adenocarcinoma patients was classified and correlations with MVC were evaluated. The mean MVC of 24 pancreatic adenocarcinomas (31.6 +/- 11.1) was actually lower than that of HCCs (91.6) or ICTs (56.4). The diversity is temperate as compared with that of other adenocarcinomas, i.e., 42.9 in gastric carcinomas, 35.6 in colon carcinomas and 32.5 in IPMT. MVC in non-cancerous areas were significantly higher in the pancreas (112.8) than in the stomach (29.6) or colon (26.3). MVC ratios of the cancerous area to the non-cancerous area were significantly lower in the pancreas (0.2818 +/- 0.100) than in the stomach (1.569 +/- 0.526, p<0.001) or the colon (1.423 +/- 0.493, p<0.001). MVC were higher in diffuse hepatic metastasis patients (36.0) than in limited metastasis patients (25.7). In conclusion, MVC in pancreatic adenocarcinoma revealed vascular volume to actually be lower than that of hypervascular tumors. We believe, however, that this hypovascularity is due mainly to contrast with the hyper-vascular non-cancerous pancreas, since MVC in the cancerous area itself was at the same level as in other adenocarcinomas. In addition, we revealed MVC to be of value for predicting the extent of liver metastasis.